Jordan Grafman
NINDS
Bethesda, MD
The prefrontal cortex (PFC) has been implicated in a
variety of “higher” cognitive functions – language, abstract reasoning, problem
solving, social interactions, and planning. A number of theories have been
proposed for how the PFC might mediate these functions.
This course will focus on strategies you can use to think about how to
characterize the functions of the human prefrontal cortex. I will briefly
discuss the evolution, development, and neuroanatomy of the human prefrontal
cortex followed by a summary of the advantages and pitfalls of studying
normal volunteers and patients using both functional neuroimaging and the
lesion approach. Then, I will present the five criteria that a theory
should meet if it is to provide a useful framework for understanding the
functions of the prefrontal cortex. The bulk of the course will be an overview
of some of the key PFC theories and my determination of how well they meet
this criteria. The criteria by which the theories will be assessed
are: (1) specification of the type of information stored in the prefrontal
cortex; (2) consistency with our knowledge of stimulus representation in
the brain; (3) consistency with what is known of the evolutionary development
of the prefrontal cortex; (4) ability to test the model and hence to verify
or invalidate it; and (5) consistency of the model with available experimental
data. A theory may take a processing approach – i.e., specify computational
procedures that are performed by the prefrontal cortex in order to manipulate
information stored elsewhere in the brain. Alternatively, a theory may take
a representational approach – i.e., specify the type of information stored
in memories in the PFC. Finally, a theory might have components of both the
processing and representational viewpoints.
I will discuss the main claims of each key theory and review data addressing
these claims. You will learn that the models meet the above criteria to varying
degrees. While all models are supported to some extent by the available cognitive
neuroscience data, not all of the models address all of the available data.
In particular, many researchers rely almost solely on functional neuroimaging
data ignoring other sources of evidence (e.g., lesion studies). With
respect to specific theories, without modification, no single theory of PFC
function appears to explain all of the available data.
Finally, I will argue that the representational approach seems to be more
consistent with the above criteria than does the processing approach.
I will show how the representational approach forces a more detailed specification
of a theory and thus enables specific hypothesis testing and is the most
parsimonious way to explore the nature of knowledge stored in the human PFC.